IBM Tivoli Software IBM Tivoli Software

[ Bottom of Page | Previous Page | Next Page | Contents | Index ]


Events and indications

An indication occurs when the state of a given resource meets defined criteria. By itself, an indication does not trigger any specific action. Indications are aggregated to become an event.

An event is a notification of change in the status of a resource. The event notifies the system administrator about the state of a specific resource. Events can trigger an action, and, if enabled, send a notification to the Tivoli Enterprise Console server or to Tivoli Business Systems Manager. When you define an event, you also specify whether these indications must be consecutive, or whether the sequence can be interrupted by one or more monitoring cycles that do not register any indication (holes).

When you define the number of holes, you specify an acceptable and consecutive number of cycles in which no indication occurs. If this number is exceeded, the aggregation rule breaks and accumulation of indications restarts at zero.

Attributes

Attributes can be string or numeric values, and can be chosen according to the information that is collected by the resource model. You can associate relevant attributes to an event.

A key attribute is the most representative attribute and uniquely identifies the event to which it refers. Key attributes play a fundamental role in the event aggregation process.

For example, an event might indicate that disk space is not sufficient. By specifying attributes, such as available disk space, you can generate a more precise indication of the problem.

When selecting an attribute name, choose a name that does not match an attribute name in the base Tivoli Enterprise Console event slot (severity names can match, however). Duplicate attribute names generate an error when the resource model is imported to IBM Tivoli Monitoring.

Actions

You can configure one or more recovery actions to correct a monitored issue in your environment. When configured, these actions automatically trigger when a specific event occurs and typically restore a satisfactory system service level. The system provides notification when the action successfuly completes.

You can associate two types of actions with an event: programs and CIM methods.

Programs
On all systems, you can run a program as an action. You can also indicate a set of arguments (for example, environment variables) that are associated with the execution of the program.
CIM methods
On Windows-based resource models, you can use a CIM class method as an action. In CIM, each class has attributes that characterize the object and a set of methods that are actions related to that object. Actions invoke methods that are provided by WMI. An action can invoke one method at a time.

When you select a class within the IBM Tivoli Monitoring Resource Model Builder, the related methods that are defined by WMI are automatically displayed. A method can be static or non-static. A static method is defined with respect to a class. A non-static method is defined with respect to an instance.

For example, if you are monitoring Windows services, and an event generates when a service stops, you can associate the event with a recovery action that restarts the interrupted service.

You can implement a CIM method within the Java IBM Tivoli Monitoring Engine (Unix). The Instrumentation Library Type (ILT) needs to support the InvokeMethod ILT public operation.


[ Top of Page | Previous Page | Next Page | Contents | Index ]